The Egyptian Journal of Remote Sensing and Space Sciences (2012) 15, 135–141

National Authority for Remote Sensing and Space Sciences The Egyptian Journal of Remote Sensing and Space Sciences

www.elsevier.com/locate/ejrs www.sciencedirect.com

Research Paper Techniques of Remote Sensing and GIS for flood monitoring and damage assessment: A case study of province,

Mateeul Haq *, Memon Akhtar, Sher Muhammad, Siddiqi Paras, Jillani Rahmatullah

Pakistan Space and Upper Atmosphere Research Commission (SUPARCO), P.O. Box 8402, 75270, Pakistan

Received 22 February 2012; revised 14 May 2012; accepted 26 July 2012 Available online 24 September 2012

KEYWORDS Abstract Remote Sensing has made substantial contribution in flood monitoring and damage Remote Sensing & GIS; assessment that leads the disaster management authorities to contribute significantly. In this paper, Flood monitoring; techniques for mapping flood extent and assessing flood damages have been developed which can be Damage assessment; served as a guideline for Remote Sensing (RS) and Geographical Information System (GIS) oper- Temporal resolution ations to improve the efficiency of flood disaster monitoring and management. High temporal resolution played a major role in Remote Sensing data for flood monitoring to encounter the cloud cover. In this regard, MODIS Aqua and Terra images of Sindh province in Pakistan were acquired during the flood events and used as the main input to assess the damages with the help of GIS analysis tools. The information derived was very essential and valuable for immediate response and rehabilitation. Ó 2012 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V. All rights reserved.

1. Introduction social damages than any other natural phenomenon (DMSG, 2001). The flooding has been affecting the entire country inco- Floods are among the most devastating natural hazards in the herently since 1928 during the monsoon period between July world, widely distributed leading to significant economic and and September and the country has experienced major floods during the years 1928, 1929, 1955, 1973, 1976, 1980, 1988, 1992 and 2010 (Solheim et al., 2001; Natalia Kussul et al., * Corresponding author. Tel.: +92 2134690785; fax: +92 2008). Since 1947–2008, floods in the Basin in 2134644928. Pakistan have claimed more than 7,000 lives; inundated zone E-mail address: [email protected] (M. Haq). was 7.7 million acres round and caused massive infrastructure Peer review under responsibility of National Authority for Remote and crop losses (Muhammad Shahzad Sardar et al., 2008). The Sensing and Space Sciences. flood in 2010 began in late July inundated Pakistan with approximately 60,000 km2, affected 15–20 million people along with the death toll close to 2000 (Mateeul-Haq et al., Production and hosting by Elsevier 2010; Wikipedia, 2010). In the recent year large-scale heavy

1110-9823 Ó 2012 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ejrs.2012.07.002 136 M. Haq et al.

Figure 1 The location of study area. rains have been observed in Sindh leading to huge economic roads, railway tracks, vegetated areas and other land use/ losses, destruction of ecological resources, food shortages land cover categories. and starvation of million people.  Moderate Resolution Imaging Spectro-radiometer In the age of modern technology, the integration of infor- (MODIS) onboard TERRA and AQUA images of 250 m mation extracted through Geographical Information System resolution comprising bands 7, 2 and 1 were acquired and (GIS) and Remote Sensing (RS) with other datasets provides used as the principal input to map the flood affected areas. tremendous potential for identification, monitoring and assess- The MODIS images of the study area are provided by NASA ment of flood disaster (Pradhan et al., 2009; Pradhan and free of charge at the location: http://lance-modis.eosdis.nasa. Shafie, 2009). This paper describes the procedure for mapping gov/imagery/subsets/?project=other&subset=Pakistan inundated areas using MODIS data during Sindh flood 2011 to  District wise population density was used to estimate the estimate the affected land use/land cover types and the number affected people and the data are provided by Federal of people. This would be very useful information for disaster Bureau of Statistics, Pakistan (Statistical Bulletin, 2011). response and mitigation.  The Spot 5 images had been mosaicked and used for the verification of flood. 2. Study area and climate  ENVI 4.5 and ArcGIS 9.3 softwares were used for process- ing and analysis. The study area is located in southern part of the country stretching about 579 km from north to south and 442 km from east to west with an area of 140915 km2 (See Fig. 1). 4. Methodology The annual rainfall of Sindh averages about 6–7 inches (15– 18 cm) falling mainly during July and August. The study area MODIS images were acquired from the NASA’s website: lies between two monsoon––thesouthern monsoon from the In- http://lance-modis.eosdis.nasa.gov/imagery/subsets/?project= dian Ocean and the northeast or retreating monsoon, deflected other&subset=Pakistan and a standard supervised maximum towards it by the Himalayan mountains. The region’s scarcity of likelihood classification of the images covering Sindh province water is compensated by the inundation of the Indus twice a was carried out and the required inundated class was con- year, caused by the spring and summer melting of Himalayan verted into a shape file. Further editing of the images was car- snow and by rainfall in the monsoon season (Wikipedia, 2012). ried out by visual interpretation (stagnant water has dark blue reflectance and flowing water has light blue color) of the inun- 3. Data and softwares used dated areas using expert knowledge. The topographic maps were then intersected with accumulated shape file of the inun- dated area to extract different types of info-layers for damage  Topographic maps had been used to extract different types assessment. Flow chart of the methodology adopted in this of info-layers: administrative boundaries, rivers, lakes, study is shown in Fig. 2. Techniques of Remote Sensing and GIS for flood monitoring and damage assessment 137

Figure 2 Flow chart of methodology.

Cumulative Flood Extent 15/09/2011 4500.00 4000.00 3500.00 2 3000.00 2500.00 2000.00

Area / Km 1500.00 1000.00 500.00 0.00 Dadu Badin Thatta Ghotki Sukkur Matiari Larkana Sanghar Khairpur Umerkot Shikarpur Jamshoro Kashmore Jacobabad Hyderabad Tharparkar Mirpurkhas Shahdadkot Nousheroferoz Tando Allahyar Tando M. Khan Shaheed Benazirabad

Figure 3 Cumulative flood extent of different districts.

5. Results and discussion different districts on different dates while Figs. 4a–d show the cumulated inundated area in Sindh. During floods, timely and detailed situation reports are required The flood was originated mostly because of rainfall and ob- by the disaster management authorities to locate and identify served the highest ever recorded monsoon rain in Sindh started the affected areas and to implement the corresponding damage from Aug 11, 2011 to Sept 14, 2011. So the inundated area in- mitigation; this is the most delicate management category since creased respectively with rainfall and ceased on Sept 15, 2011 effectively with the stop of rainfall. After Sept 15, 2011 the it involves rescue operations and the safety of people and prop- 2 erty (Jeyaseelan, 2004). In this regard, cumulative and temporal inundated area was reduced with the rate of 167 km /day in flood extent maps were prepared. It was found that severe flood average as the following graph shows the recession. Table 1 occurred in Badin and inundated it by 3820.39 km2, Mirpur- shows the rainfall (in mm) data (See Fig. 5). khas by 1836.26 km2, Jacobabad by 1352.32 km2, Shahdadkot by 1597.50 km2, Dadu by 1887.57 km2 and Sanghar by 5.1. Damage assessment 2494.18 km2, in cumulative. Furthermore the above mentioned districts contributed 61% of the total inundated area among 23 The total cumulative flood extent was found to be 21,201 km2 districts in Sindh. Fig. 3 shows the cumulative inundated area of affected 5.88 million people, 5329 settlements, 1500.44 km of 138 M. Haq et al.

Figure 4a (a) MODIS image showing the inundated area on Aug 20, 2011.

Figure 4b (b) MODIS image showing the inundated area on Aug 31, 2011. Techniques of Remote Sensing and GIS for flood monitoring and damage assessment 139

Figure 4c (c) MODIS image showing the inundated area on Sept 04, 2011.

Figure 4d (d) MODIS image showing the inundated area on Sept 15, 2011. 140 M. Haq et al.

Table 1 Data had been taken from website of Pakistan Meteorological Department (Pakmet, 2011). Date Aug-11 Sep-11 District 11 12 30 31 1 2 4 7 8 9 13 14 Total Badin 148 147 22 7 15 24 19 73 20 11 60 0 546 Umerkoat 129 7 19 57 6 5 11 55 84 8 56 0 437 Hyderabad 104 23 4 15 2 27 0 2 12 12 153 0 354 Jacobabad 0 0 0 4 99 0 0 0 0 0 ** 82 185 Larkana 0 0 0 71 40 37 0 16 8 0 ** 21 193 Therparkar 291 60 42 0 3 43 12 225 137 100 57 0 970 S. Benazeerabad 5 108 55 70 19 67 70 25 37 27 75 6 564 Nausheroferoz 1 0 8 238 10 57 0 0 18 2 42 0 376 Sukkur 0 0 0 66 33 28 4 4 0 0 4 12 151 Thatta 58 7 43 0 0 6 0 9 6 0 73 4 206 Dadu 0 0 0 130 2 48 0 0 66 9 115 108 478 Mirpurkhas 120 125 12 5 4 50 50 1 166 122 190 0 845

Receding Inundated Area 18000 16000 14000 )

2 12000 10000 8000

Area (Km 6000 4000 2000 0

Figure 5 Graph showing the recession of the inundated area from Sept 15, 2011.

Table 2 Statistical analysis of flood-affected land cover/land use categories based on MODIS data. S. No District name Affected Affected Affected agricultural Affected Affected Affected settlements people land/Km2 forests/Km2 roads/km railway/km 1 Badin 967 853993 1219.05 12.64 146.09 14.65 2 Dadu 261 373386 1483.94 12.14 168.08 31.76 3 Ghotki 237 189753 880.43 83.04 29.04 14.55 4 Hyderabad 7 80951 25.07 5.67 5.96 4.08 5 Jacobabad 349 520535 1371.07 0.00 71.81 58.64 6 Jamshoro 59 42096 327.29 47.62 43.14 42.37 7 Kashmore 194 243371 690.38 28.69 31.04 11.63 8 Khairpur 144 49886 314.01 34.29 36.09 19.03 9 Larkana 68 158712 219.36 1.65 23.64 12.86 10 Matiari 24 91215 91.45 85.09 11.21 4.00 11 Mirpurkhas 544 869569 1852.84 4.18 196.12 18.18 12 Nousheroferoz 243 323765 630.46 32.19 48.71 36.83 13 Sanghar 657 427972 2203.68 0.00 268.21 19.87 14 Shahdadkot 316 375152 1196.09 0.00 79.23 13.25 15 Shaheed Benazirabad 328 298067 823.27 8.82 110.17 28.60 16 Shikarpur 248 389847 807.03 19.88 38.82 23.24 17 Sukkur 49 54282 188.42 2.05 6.13 9.01 18 Tando Allahyar 75 138732 329.38 0.00 14.52 3.64 19 Tando M. Khan 83 163307 270.34 4.71 21.54 1.64 20 Tharparkar 43 13779 193.07 4.19 0.00 21 Thatta 133 83069 437.52 115.81 42.07 0.00 22 Umerkot 300 139619 886.35 0.00 104.64 14.21 Total 5329 5881056 16440.48 498.47 1500.44 382.05 Techniques of Remote Sensing and GIS for flood monitoring and damage assessment 141 road network, 382.05 km of railway tracks, 498.47 km2 of for- Jeyaseelan, A.T., 2004. Droughts & floods assessment and monitor- ests and 16440.48 km2 of agricultural land. Table 2 shows the ing using Remote Sensing and GIS. Satellite Remote Sensing and damage assessment. GIS Applications in Agricultural Meteorology, 291–313. Kussul, Natalia, Shelestov, Andrii, Skakun, Serhiy, Kravchenko, Oleksii, 2008. Data assimilation technique for flood monitoring 6. Conclusion and prediction. International Journal Information Theories & Applications 15, 76–83. Flood monitoring using satellite data proved to be an effective Mateeul-Haq, Said Rahman, Rahmatullah Jillani and Sher Moham- method to get quick and precise overview of flooded areas. In mad, 2010. Pakistan – flood 2010 monitoring using MODIS data. the study, timely and detailed analysis had been carried out In: the 9th International Workshop of the CAS-TWAS-W Forum using RS & GIS for locating and identifying flood affected (2010 CTWF) on Climate and Environmental Changes and Challenges for Developing Countries, 17–20 November, Beijing, areas along with land use/land cover features. It was found China. that this method required processed satellite images which Sardar, Muhammad Shahzad, Tahir, Muhammad Avais, Zafar, were then overlaid with population density data and land Muhammad Iqbal, 2008. Poverty in riverine areas: vulnerabilities, use/land cover maps for damage estimation. The process just social gaps and flood damages. Pakistan Journal of Life and Social required a few hours. The methodology used in this study Science 6 (1), 25–31. has the capability to carry out rapid damage assessment. Pakistan Meteorological Department. Available at . Pradhan, B., Shafie, M., 2009. Flood hazard assessment for cloud prone rainy areas in a typical tropical environment. Disaster Acknowledgements Advances 2 (2), 7–15. Pradhan, B., Pirasteh, S., Shafie, M., 2009. Maximum flood prone The work reported was carried out for the National Disaster area mapping using RADARSAT images and GIS: Kelantan river Management Authority, Pakistan and Sindh Irrigation basin. International Journal of Geoinformatics 5 (2), 11–23. Department, Pakistan. We thank to all the concerned depart- Singapore Red Cross, 2010. Pakistan Floods: the Deluge of Disaster ments that had provided us with data. – Facts & Figures as of 15 September 2010, http://en.wikipedia.org/ wiki/2010_Pakistan_floods. Solheim, I., Solbo, S., Indregard, M., Lauknes, I., 2001. User References Requirements and SAR-Solutions for Flood Mapping. In: 4th International Symposium on Retrieval of Bio- and Geophysical DMSG, 2001. The Use of Earth Observing Satellites for Hazard Parameters from SAR Data for Land Applications, Innsbruck, Support: Assessments & Scenarios. Committee on Earth Observa- Austria. tion Satellites Disaster Management Support Group, Final Report, Wikipedia, 2012. The Geography and Climate of Sindh, http:// NOAA, Dept. Commerce, USA. en.wikipedia.org/wiki/Sindh. Federal Bureau of Statistics, 2011. Pakistan, Statistical Bulletin.